CN113867909A

CN113867909A - Atomic real-time liquidation method

Info

Publication number: CN113867909A
Application number: CN202111111777.1A
Authority: CN
Inventors: 茅冬琳; 黄兆毅; 童军; 李润东; 梁诗诗; 雷利军; 黄进明; 周红烈; 沈宪阳; 黄溢桥; 段奥; 邓欢; 罗梦瑶; 嵇秉容
Original assignee: Shenzhen Qiaoxingu Information Technology Co ltd
Current assignee: Shenzhen Qiaoxingu Information Technology Co ltd
Priority date: 2021-09-23
Filing date: 2021-09-23
Publication date: 2021-12-31

Abstract

The invention relates to the technical field of financial science and technology, in particular to an atomic real-time clearing method, which comprises the following steps: step S1: dividing an atomic-level fine-grained task; step S2: generating a dependency relationship between tasks while generating the tasks; step S3: performing real-time clearing after the dependence condition is met; step S4: lightweight recovery is performed as needed. The traditional clearing mode can be analogized to a long-distance bus mode, for example, only one shift is carried out in one day, and the bus can be dispatched after all data are in order; the invention can be analogized to a private car mode, each data clearing related party has an independent task flow, each task is driven independently, and each task is orderly, so that the maximum flexibility, the minimum flow dependence, the fastest recovery speed and the minimum recovery risk are realized.

Description

Atomic real-time liquidation method

Technical Field

The invention relates to the technical field of financial science and technology, in particular to an atomic real-time clearing method.

Background

The ta (transfer agent) system is a registration system for managing customer assets by an asset management mechanism, and is mainly used for managing customer accounts, clearing customer transactions, managing customer shares, and allocating customer rights and interests.

The TA system generally undertakes the clearing tasks of a plurality of channels (china bank, agricultural bank … …) and a plurality of products (consumption fund, military fund … …), calculates the share of the customer through the business rules, and informs the channels of the result after clearing is completed. The traditional TA system clearing design logic is that clearing can be started only after data files corresponding to all products and all channels are imported into the traditional TA system clearing design logic when clearing is carried out every working day.

The clearing system used by the existing market asset management mechanism basically adopts a uniform clearing mode, clearing can be carried out only after all files are collected, all data can be backed up before clearing, and all data can be restored and then cleared again when problems occur.

This mode has the following disadvantages:

1. a large amount of time is consumed to unnecessarily wait for each other, and the overall clearing efficiency is low.

The product cannot be processed in batches and the workload of the operation is concentrated at a certain time. As product types continue to increase, the time to market for portions of the product may be relatively late, resulting in a delay in the overall system clearing time. For example, operator A has finished importing the product 1 clearance file at 15:30, and operator B has imported the product 2 clearance file at 19: 00. The unified clearing can be started only at 19:00, the clearing is finished at 20:00 of the system, and the operation A has to wait for 20:00 to take the result data for the next operation;

2. the problem is exposed late, resulting in a narrow window of treatment and recovery time.

Usually, the time window reserved for batch clearing is limited, but the traditional clearing mode consumes a lot of time to wait for each other, so that the time window reserved for subsequent problem positioning and problem solving is greatly shortened when clearing, checking and problem exposure are really started, and a larger operation risk is caused.

3. The recovery influence range is large, the speed is slow, and the risk is high.

When clearing is needed to restore the clearing result after an error occurs, the traditional mode is to uniformly restore through backup of a database. In fact, clearing errors often occur on a single product, a single channel, or even a single transaction, and database backup schemes require resetting the clearing results of the entire system, create unnecessary recovery of other data, and raise more unexpected risks.

In view of this, we propose an atomic real-time liquidation method.

Disclosure of Invention

The present invention is directed to an atomic real-time liquidation method to solve the above problems.

In order to achieve the purpose, the invention provides the following technical scheme: an atomic real-time liquidation method, comprising the steps of:

step S1: dividing an atomic fine-grained task;

step S2: generating a dependency relationship between tasks while generating the tasks;

step S3: performing real-time clearing after the dependence condition is met;

step S4: and performing lightweight recovery. The traditional liquidation mode can be analogized to a coach mode, for example: only one shift is carried out in one day, and the car can be dispatched after all data are in order; the invention can be analogized to a private car mode, each data clearing related party has an independent task flow, each task is driven independently, and each task is orderly, so that the maximum flexibility, the minimum flow dependence, the fastest recovery speed and the minimum recovery risk are realized.

Preferably, the task generation in step S1 may be manually generated by an operator, or automatically generated by the system, and a complete task flow including a task list and a sequential dependency relationship on a certain date is generated for each clearing participant; and generating a dependency relationship between tasks among different participants according to business needs.

Preferably, the task dependency in step S2 follows an on-demand dependency, minimum dependency rule, and if the input information of a task is the output of a certain task, a bar dependency is generated.

Preferably, the real-time clearing in step S3 is performed when the task satisfies all the dependent conditions, and can be immediately executed without redundant waiting.

Preferably, the lightweight recovery in step S4 includes a single-stroke recovery for implementing the clearing, a single-task recovery, and a re-clearing after implementing the clearing.

Preferably, the data source recovered in step S4 is based on both addable data and unadditive data, the addable data is a new changed data record in the form of a flow meter, and the unadditive data is a backup data record.

Compared with the prior art, the invention has the beneficial effects that:

1. the operation efficiency is improved, and the clearing speed is greatly improved (in practice, the clearing completion time is improved from about 23 points to about 20 points, and the integral speed is improved by more than 3 hours);

2. the method helps to find and solve problems more quickly (in practice, when data of a certain channel is wrong, the problem finding usually needs about 2-3 hours, and the problem finding time is shortened to be within 5 minutes after the method is adopted);

3. the data recovery operation has small influence range, high speed and low risk, the original recovery and recalculation mode in practice usually takes 1-4 hours according to the data size, and after the method is adopted, the recalculation is usually recovered by only one task, and the time is about 5-20 minutes.

Drawings

FIG. 1 is a system block diagram of the clearing method of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1, the present invention provides a technical solution: an atomic real-time liquidation method, comprising the steps of:

step S1: dividing an atomic fine-grained task;

step S3: performing real-time clearing after the dependence condition is met;

step S4: and performing lightweight recovery.

The traditional liquidation mode can be analogized to a coach mode, for example: only one shift is carried out in one day, and the car can be dispatched after all data are in order; the invention can be analogized to a private car mode, each data clearing related party has an independent task flow, each task is driven independently, and each task is orderly, so that the maximum flexibility, the minimum flow dependence, the fastest recovery speed and the minimum recovery risk are realized.

Aiming at the problems that the prior clearing scheme can not clear in batches flexibly, has long waiting time, low operation efficiency, large recovery influence range, long consumed time and the like, the scheme solves the following problems:

1. dividing the unified task with the finest granularity into a plurality of small tasks;

2. managing the dependency relationship among tasks;

3. lightweight recovery.

And generating clearing tasks according to the dimensions of products, channels and the like, dividing one large task into a plurality of small tasks, allowing the tasks to be parallel, executing or recovering each task after the task meets the requirement of dependence, and finding problems as early as possible if a single task has problems. The generation and dependency relationship of the tasks are the basis of the scheme.

The method mainly comprises the following four aspects:

1. atomic fine-grained task partitioning

The task generation of each working day can be manually generated by operators or automatically generated by a system, the task is divided into three dimensions of TA, a channel and a product, the two dimensions of the channel and the product are crossed to form a matrix, and each product and each channel have an independent task flow.

TA dimension: and (4) performing daily end tasks, uniformly performing task initialization, archiving data and the like.

Channel dimension: the clearing process of the channel visual angle mainly comprises the steps of application import, data clearing, confirmation export and the like, wherein the import and export time points of each channel are different, and the data summarization dimensions are different, for example, the data summarization process can be carried out according to the product or can summarize all the products; the main parameters are channel codes and channel clearing dates.

Third, product dimension: the clearing process of the product visual angle mainly comprises data clearing, product-level quota control and the like. The task of channel and product intersection (such as transaction clearing of a certain product in a certain channel) belongs to the product dimension and mainly participates in product codes and product clearing days.

2. Strict dependency relationships

And when the tasks are generated, the dependency relationship among the tasks can be generated at the same time, and the task dependency needs to follow the principle of dependency according to requirements and minimum dependency.

If the input information of a task is the output of a certain task, a bar dependency is generated. And generating the dependence according to the rule, and avoiding generating unnecessary dependence.

The task dependency generation follows the following rules:

if: the output result of the execution of the task A is part of the input parameters of the execution of the task B

Then: task B depends on task A

For example: the output of the import application task is application data, and the transaction clearing is clearing according to the application data, so the transaction clearing needs to depend on the import application task.

3. Real-time liquidation

The task can be executed immediately after all the dependent conditions are met without redundant waiting.

4. Lightweight recovery

If the data generated in the task execution is addable, the data is added in the form of a flow meter, for example: and if the balance of the customer changes, adding a balance change record, wherein the change value can be a positive number or a negative number, and when the balance needs to be recovered, deleting the corresponding change record.

And for the data which is not added and needs to be updated, a backup mode is adopted to backup the data before change. And when in recovery, the recovery is carried out according to the backup.

Based on the two schemes, the single-stroke recovery of clearing can be realized. If a problem occurs in the clearing process, the error data single-stroke recovery can be selected, and all data can be recovered according to the task dimension for clearing again.

When recovering, all tasks depending on the task are found in sequence, and the recovery is carried out one by one in a recursive manner, wherein the logic of the recovery is as follows:

RECOVER (task)

1. Querying tasks that depend on the task and have already been executed

2. If the query result is not null, continuing to execute:

RECOVER (task of query)

3. Performing recovery processing

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims

1. An atomic real-time liquidation method is characterized in that: the real-time liquidation method comprises the following steps:

step S1: dividing an atomic-level fine-grained task;

step S3: performing real-time clearing after the dependence condition is met;

step S4: lightweight recovery is performed as needed.

2. An atomic real-time liquidation method according to claim 1, wherein: the task generation in step S1 may be triggered manually by an operator, or may be automatically generated by the system, and a complete task flow on a certain date is generated for each clearing participant; and generating a dependency relationship between tasks among different participants according to business needs.

3. An atomic real-time liquidation method according to claim 1, wherein: in the step S2, the task dependency needs to follow the on-demand dependency, the minimum dependency principle, and if the input information of the B task is the output of the a task, the dependency relationship of the B task on the a task is generated.

4. An atomic real-time liquidation method according to claim 1, wherein: the real-time clearing in step S3 is performed on the condition that the task satisfies all the dependency conditions, and can be immediately executed without waiting unnecessarily.

5. An atomic real-time liquidation method according to claim 1, wherein: the lightweight recovery in step S4 includes two aspects of single-stroke recovery for implementing liquidation, single-task recovery, and recleaning after implementing liquidation.

6. An atomic real-time liquidation method according to claim 5, wherein: the data source recovered in step S4 is based on both the addable data and the unadditive data, the addable data is a new change data record in the form of a flow meter, and the unadditive data is a backup data record.